Inertia Meets Existential Crisis

We have not one but two things to talk about this weekend.

Broadly they are

1. Is AI producing value?

2. What to pay for it?

AI is amazing but it delivers results for you, only if you can tear how work is done. If you have not had a townhall meeting to say “All titles are relics” or “everyone is a manager now,” then you simply have not started the re-engineering.

Secondly, your friendly SaaS vendor and you are running loops around each other. They want to prove worth by concocting a pricing model that pegs on outcomes. You would never want to open the kimono and share what dollars you have saved. Resist fancy pricing models that are hard to evaluate. Don’t sign multi-year contracts. Every line of code is getting 10 times cheaper than it was last month. Your SaaS vendor is only needed if your workflow is complex, the future states need a vendor that thinks ahead, and compliance/assurance are not burdens you can take in-house.

Let’s talk about the first thing now: Productivity Unlock

80% of firms report no productivity gains from AI. The interesting question is why.

An NBER study published this month surveyed nearly 6,000 executives across four countries. Over 80% of firms report no measurable impact on employment or productivity despite widespread AI adoption. PwC's 2026 CEO survey: only 12% of CEOs saw both cost and revenue benefits from AI. 56% reported zero significant financial benefit. Executives use AI about 1.5 hours per week on average.

This is Robert Solow's paradox, back from the dead. In 1987, he observed that computers were everywhere except in the productivity statistics. It took fifteen years before that investment paid off. We are in the same moment.

But Solow's paradox had an answer, and the answer was that organizations eventually redesigned how work was done. That hasn't happened yet with AI. We see it in every client engagement.

Where the time goes

Workday's 2026 research surfaced a number that should worry every operations leader: 37 to 40% of the time "saved" by AI gets consumed reviewing, correcting, and verifying AI output. It is fair to call this "workslop."

Your team adopts a tool that promises to save four hours a week. They gain those four hours. Then they spend ninety minutes checking and fixing what the tool produced. Net gain: two and a half hours, and that's the generous estimate. The cognitive load of constant verification doesn't show up in any time-saved metric.

We see the same pattern across industries. Companies layer AI on top of existing workflows, then layer human review on top of the AI. The same approvals, the same handoffs, the same bottlenecks. One more step in the middle.

The NBER data tells the story: Of the 70% of firms actively using AI, over 80% report no productivity impact. That gap is not a mystery. Brynjolfsson, Rock and Syverson have documented the pattern across every general-purpose technology: the returns only show up after organizations restructure around the capability. Electricity took thirty years. PCs took fifteen. Depending on who you ask, we are maybe in year three with AI, and most companies haven't changed a single approval chain or reporting structure.

Nobody in the vendor ecosystem has an incentive to close this gap. Your SaaS vendor wants you to add their AI feature. Your cloud provider wants you to consume more compute. Your consulting firm wants to bill you for an AI strategy. The hard, unglamorous work of redesigning how an organization operates? 

That's on you.

Rethink org design now. If you have gone ahead with AI and you have not redesigned the org, you are wasting money.

What to pay for AI?

The conventional wisdom says SaaS should shift from per-seat pricing to outcome-based pricing. Charge for value delivered, not for humans who log in. It sounds elegant.

It is also unworkable.

The entire outcome-based pricing thesis assumes value can be measured, attributed, and agreed upon by both parties. In practice, your customer will never voluntarily tell you how much money your tool generated for them. They will downplay the impact to negotiate a lower price. They will attribute results to their own team's effort. They will restructure how they account for gains to minimize what they owe you. This is how procurement works. Every vendor negotiation is an information asymmetry game. Outcome-based pricing hands the information advantage entirely to the buyer.

The only outcome model that works is a share of terminal value that comes from shared equity. A customer has no incentive to disclose the value your tool generated for them. The per-seat model was never about capturing value. It was about simplicity and predictability on both sides. Replacing it with "pay me based on outcomes" doesn't solve SaaS's pricing crisis. It creates a measurement crisis and a trust crisis on top.

So what should you do or not do?

1. Don’t be your SaaS vendor’s pricing experiment. Resist convoluted pricing models

2. Don’t sign multi-year contracts with automatic inflation clauses

3. Don’t buy SaaS just for features

4. Don’t fire your SaaS vendors without assessing if they also deliver assurance, compliance, and future-readiness

Two things failing at once

So here's where we are. Two mainstream narratives, all struggling under their own weight.

"AI will boost productivity" 80% of firms say it hasn't. The time saved is getting eaten by verification loops and unchanged workflows.

"Shift to outcome-based pricing" Information asymmetry makes it unenforceable. The only honest outcome alignment is shared equity. Only a few SaaS vendors can charge a premium and only when they bundle software with assurance. Most think they bring that value. You need to judge if your CRM and scheduling software tools are just features or a layer that evolves as your workflows in their area evolve in a complex way and are subject to key risks and compliance burdens. You know the answer already.

Ironically, your inertia prevents you from re-engineering your organization. Your vendor’s existential crisis prevents them from accepting that they are a commodity. 

Solow's paradox resolved itself eventually. But it resolved because companies rebuilt their organizations around what computers made possible. We need the same willingness now, and the path probably looks less like buying new tools and more like rebuilding how the work itself is structured.

Enjoy your Sunday!

Top reads